正交设计的E占优策略求解高维多目标优化问题研究

在实际应用中,传统多目标演化算法面临着高维多目标优化问题。针对这一缺陷,提出正交E占优(Orthogo-nality E-dominant,OE)策略。在OE策略的理论优越性设计的基础上,改进了当前5种具有代表性的演化多目标优化算法。改进前后的算法求解DTLZ1-6(20)测试问题的数值对比试验显示,OE策略改进后的算法在不同程度上提高了算法求解高维多目标优化问题的效果,从而证实了OE策略对演化多目标优化算法改进的有效性。

Research of Orthogonal E-dominant(OE) Strategy to Solve Large-dimensional Objective Optimization Problems

This paper proposed orthogonal E-dominant(OE) strategy to solve this defect of classical multi-objective optimization algorithms.This strategy is used to improve the current evolution multi-objective optimization algorithm,to make the algorithm to have not adequate choice pressure in individuals and global search.Based on the theory of the OE strategy’s superiority,we improved the current representative evolutionary multi-objective optimization algorithms.Compared with the classic algorithm in solving DTLZ1-6(20),the OE strategy improves the algorithm in varying degrees to solve large-dimensional multi-objective optimization problems,reaches the goal of solving large-dimensional multi-objective optimization problems,and is effective to improve the multi-objective evolutionary algorithm.